Skip to content

GitLab

Commit 2186e9f2 authored by vishnubanna's avatar vishnubanna
Browse files

Merge branch 'yolo' of https://github.com/PurdueCAM2Project/tf-models into yolo

parents 669807e6 da0d0d27
Show whitespace changes
Inline Side-by-side
Showing with 73 additions and 74 deletions
official/vision/beta/projects/yolo/modeling/layers/nn_blocks.py
View file @ 2186e9f2
...@@ -33,20 +33,20 @@ class DarkConv(ks.layers.Layer): ...@@ -33,20 +33,20 @@ class DarkConv(ks.layers.Layer):
strides: integer of tuple how much to move the kernel after each kernel use strides: integer of tuple how much to move the kernel after each kernel use
padding: string 'valid' or 'same', if same, then pad the image, else do not padding: string 'valid' or 'same', if same, then pad the image, else do not
dialtion_rate: tuple to indicate how much to modulate kernel weights and dialtion_rate: tuple to indicate how much to modulate kernel weights and
the how many pixels ina featur map to skip how many pixels in a feature map to skip
use_bias: boolean to indicate wither to use bias in convolution layer use_bias: boolean to indicate whether to use bias in convolution layer
kernel_initializer: string to indicate which function to use to initialize weigths kernel_initializer: string to indicate which function to use to initialize weights
bias_initializer: string to indicate which function to use to initialize bias bias_initializer: string to indicate which function to use to initialize bias
kernel_regularizer: string to indicate which function to use to regularizer weights kernel_regularizer: string to indicate which function to use to regularizer weights
bias_regularizer: string to indicate which function to use to regularizer bias bias_regularizer: string to indicate which function to use to regularizer bias
group_id: integer for which group of features to pass through the conv. group_id: integer for which group of features to pass through the conv.
groups: integer for how many splits there should be in the convolution feature stack input groups: integer for how many splits there should be in the convolution feature stack input
grouping_only: skip the convolution and only return the group of featuresindicated by grouping_only grouping_only: skip the convolution and only return the group of features indicated by grouping_only
use_bn: boolean for wether to use batchnormalization use_bn: boolean for whether to use batch normalization
use_sync_bn: boolean for wether sync batch normalization statistics use_sync_bn: boolean for whether sync batch normalization statistics
of all batch norm layers to the models global statistics (across all input batches) of all batch norm layers to the models global statistics (across all input batches)
norm_moment: float for moment to use for batchnorm norm_moment: float for moment to use for batch normalization
norm_epsilon: float for batchnorm epsilon norm_epsilon: float for batch normalization epsilon
activation: string or None for activation function to use in layer, activation: string or None for activation function to use in layer,
if None activation is replaced by linear if None activation is replaced by linear
leaky_alpha: float to use as alpha if activation function is leaky leaky_alpha: float to use as alpha if activation function is leaky
...@@ -91,7 +91,7 @@ class DarkConv(ks.layers.Layer): ...@@ -91,7 +91,7 @@ class DarkConv(ks.layers.Layer):
self._kernel_regularizer = kernel_regularizer self._kernel_regularizer = kernel_regularizer
self._bias_regularizer = bias_regularizer self._bias_regularizer = bias_regularizer
# batchnorm params # batch normalization params
self._use_bn = use_bn self._use_bn = use_bn
if self._use_bn: if self._use_bn:
self._use_bias = False self._use_bias = False
...@@ -137,7 +137,6 @@ class DarkConv(ks.layers.Layer): ...@@ -137,7 +137,6 @@ class DarkConv(ks.layers.Layer):
kernel_regularizer=self._kernel_regularizer, kernel_regularizer=self._kernel_regularizer,
bias_regularizer=self._bias_regularizer) bias_regularizer=self._bias_regularizer)
#self.conv =tf.nn.convolution(filters=self._filters, strides=self._strides, padding=self._padding
if self._use_bn: if self._use_bn:
if self._use_sync_bn: if self._use_sync_bn:
self.bn = tf.keras.layers.experimental.SyncBatchNormalization( self.bn = tf.keras.layers.experimental.SyncBatchNormalization(
...@@ -207,18 +206,18 @@ class DarkTiny(ks.layers.Layer): ...@@ -207,18 +206,18 @@ class DarkTiny(ks.layers.Layer):
Args: Args:
filters: integer for output depth, or the number of features to learn filters: integer for output depth, or the number of features to learn
use_bias: boolean to indicate wither to use bias in convolution layer use_bias: boolean to indicate whether to use bias in convolution layer
kernel_initializer: string to indicate which function to use to initialize weigths kernel_initializer: string to indicate which function to use to initialize weights
bias_initializer: string to indicate which function to use to initialize bias bias_initializer: string to indicate which function to use to initialize bias
kernel_regularizer: string to indicate which function to use to regularizer weights kernel_regularizer: string to indicate which function to use to regularize weights
bias_regularizer: string to indicate which function to use to regularizer bias bias_regularizer: string to indicate which function to use to regularize bias
use_bn: boolean for wether to use batchnormalization use_bn: boolean for whether to use batch normalization
use_sync_bn: boolean for wether sync batch normalization statistics use_sync_bn: boolean for whether to sync batch normalization statistics
of all batch norm layers to the models global statistics (across all input batches) of all batch norm layers to the models' global statistics (across all input batches)
group_id: integer for which group of features to pass through the csp tiny stack. group_id: integer for which group of features to pass through the csp tiny stack.
groups: integer for how many splits there should be in the convolution feature stack output groups: integer for how many splits there should be in the convolution feature stack output
norm_moment: float for moment to use for batchnorm norm_moment: float for moment to use for batch normalization
norm_epsilon: float for batchnorm epsilon norm_epsilon: float for batch normalization epsilon
activation: string or None for activation function to use in layer, activation: string or None for activation function to use in layer,
if None activation is replaced by linear if None activation is replaced by linear
**kwargs: Keyword Arguments **kwargs: Keyword Arguments
...@@ -314,22 +313,22 @@ class DarkResidual(ks.layers.Layer): ...@@ -314,22 +313,22 @@ class DarkResidual(ks.layers.Layer):
Args: Args:
filters: integer for output depth, or the number of features to learn filters: integer for output depth, or the number of features to learn
use_bias: boolean to indicate wither to use bias in convolution layer use_bias: boolean to indicate whether to use bias in convolution layer
kernel_initializer: string to indicate which function to use to initialize weigths kernel_initializer: string to indicate which function to use to initialize weights
bias_initializer: string to indicate which function to use to initialize bias bias_initializer: string to indicate which function to use to initialize bias
kernel_regularizer: string to indicate which function to use to regularizer weights kernel_regularizer: string to indicate which function to use to regularizer weights
bias_regularizer: string to indicate which function to use to regularizer bias bias_regularizer: string to indicate which function to use to regularizer bias
use_bn: boolean for wether to use batchnormalization use_bn: boolean for whether to use batch normalization
use_sync_bn: boolean for wether sync batch normalization statistics use_sync_bn: boolean for whether sync batch normalization statistics
of all batch norm layers to the models global statistics (across all input batches) of all batch norm layers to the models global statistics (across all input batches)
norm_moment: float for moment to use for batchnorm norm_moment: float for moment to use for batch normalization
norm_epsilon: float for batchnorm epsilon norm_epsilon: float for batch normalization epsilon
conv_activation: string or None for activation function to use in layer, conv_activation: string or None for activation function to use in layer,
if None activation is replaced by linear if None activation is replaced by linear
leaky_alpha: float to use as alpha if activation function is leaky leaky_alpha: float to use as alpha if activation function is leaky
sc_activation: string for activation function to use in layer sc_activation: string for activation function to use in layer
downsample: boolean for if image input is larger than layer output, set downsample to True downsample: boolean for if image input is larger than layer output, set downsample to True
so the dimentions are forced to match so the dimensions are forced to match
**kwargs: Keyword Arguments **kwargs: Keyword Arguments
''' '''
...@@ -472,24 +471,24 @@ class CSPTiny(ks.layers.Layer): ...@@ -472,24 +471,24 @@ class CSPTiny(ks.layers.Layer):
Args: Args:
filters: integer for output depth, or the number of features to learn filters: integer for output depth, or the number of features to learn
use_bias: boolean to indicate wither to use bias in convolution layer use_bias: boolean to indicate whether to use bias in convolution layer
kernel_initializer: string to indicate which function to use to initialize weigths kernel_initializer: string to indicate which function to use to initialize weights
bias_initializer: string to indicate which function to use to initialize bias bias_initializer: string to indicate which function to use to initialize bias
use_bn: boolean for wether to use batchnormalization use_bn: boolean for whether to use batch normalization
kernel_regularizer: string to indicate which function to use to regularizer weights kernel_regularizer: string to indicate which function to use to regularizer weights
bias_regularizer: string to indicate which function to use to regularizer bias bias_regularizer: string to indicate which function to use to regularizer bias
use_sync_bn: boolean for wether sync batch normalization statistics use_sync_bn: boolean for whether sync batch normalization statistics
of all batch norm layers to the models global statistics (across all input batches) of all batch norm layers to the models global statistics (across all input batches)
group_id: integer for which group of features to pass through the csp tiny stack. group_id: integer for which group of features to pass through the csp tiny stack.
groups: integer for how many splits there should be in the convolution feature stack output groups: integer for how many splits there should be in the convolution feature stack output
norm_moment: float for moment to use for batchnorm norm_moment: float for moment to use for batch normalization
norm_epsilon: float for batchnorm epsilon norm_epsilon: float for batch normalization epsilon
conv_activation: string or None for activation function to use in layer, conv_activation: string or None for activation function to use in layer,
if None activation is replaced by linear if None activation is replaced by linear
leaky_alpha: float to use as alpha if activation function is leaky leaky_alpha: float to use as alpha if activation function is leaky
sc_activation: string for activation function to use in layer sc_activation: string for activation function to use in layer
downsample: boolean for if image input is larger than layer output, set downsample to True downsample: boolean for if image input is larger than layer output, set downsample to True
so the dimentions are forced to match so the dimensions are forced to match
**kwargs: Keyword Arguments **kwargs: Keyword Arguments
""" """
def __init__( def __init__(
...@@ -665,11 +664,11 @@ class CSPDownSample(ks.layers.Layer): ...@@ -665,11 +664,11 @@ class CSPDownSample(ks.layers.Layer):
bias_initializer: string to indicate which function to use to initialize bias bias_initializer: string to indicate which function to use to initialize bias
kernel_regularizer: string to indicate which function to use to regularizer weights kernel_regularizer: string to indicate which function to use to regularizer weights
bias_regularizer: string to indicate which function to use to regularizer bias bias_regularizer: string to indicate which function to use to regularizer bias
use_bn: boolean for wether to use batchnormalization use_bn: boolean for whether to use batch normalization
use_sync_bn: boolean for wether sync batch normalization statistics use_sync_bn: boolean for whether sync batch normalization statistics
of all batch norm layers to the models global statistics (across all input batches) of all batch norm layers to the models global statistics (across all input batches)
norm_moment: float for moment to use for batchnorm norm_moment: float for moment to use for batch normalization
norm_epsilon: float for batchnorm epsilon norm_epsilon: float for batch normalization epsilon
**kwargs: Keyword Arguments **kwargs: Keyword Arguments
""" """
def __init__( def __init__(
...@@ -767,11 +766,11 @@ class CSPConnect(ks.layers.Layer): ...@@ -767,11 +766,11 @@ class CSPConnect(ks.layers.Layer):
bias_initializer: string to indicate which function to use to initialize bias bias_initializer: string to indicate which function to use to initialize bias
kernel_regularizer: string to indicate which function to use to regularizer weights kernel_regularizer: string to indicate which function to use to regularizer weights
bias_regularizer: string to indicate which function to use to regularizer bias bias_regularizer: string to indicate which function to use to regularizer bias
use_bn: boolean for wether to use batchnormalization use_bn: boolean for whether to use batch normalization
use_sync_bn: boolean for wether sync batch normalization statistics use_sync_bn: boolean for whether sync batch normalization statistics
of all batch norm layers to the models global statistics (across all input batches) of all batch norm layers to the models global statistics (across all input batches)
norm_moment: float for moment to use for batchnorm norm_moment: float for moment to use for batch normalization
norm_epsilon: float for batchnorm epsilon norm_epsilon: float for batch normalization epsilon
**kwargs: Keyword Arguments **kwargs: Keyword Arguments
""" """
def __init__( def __init__(
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment